TRANSMISSION WITH NEGATIVE GRADIENTS NEAR SURFACE 



131 



•to 



1000 



RANGE m vanos 



2000 



3000 



4000 



Figure 50. Individual anomalies for Di between 20 and 40 feet. 



due to calibration do not affect the accuracy of the 

 attenuation coefficient determined by fitting a 

 straight hne to the observed anomalies. With such a 

 scatter, the uncertainties in the attenuation coef- 

 ficient a can be very substantial for those runs in 

 which the length of run was short. However, it does 

 not seem likely that observational error can account 

 for all of the scatter shown in Figures 47 and 48. 



The attenuation coefficients shown in Figures 47 

 and 48 should probably not be used for estimating 

 transmitted sound intensities when temperature gra- 

 dients are present in the top 30 ft. The results are 

 valid on the average when the transmission anomaly 

 graph is a straight line, but unfortunately there is no 



way of predicting whether or not the measured sound 

 intensities will yield a straight line, except when the 

 top 30 ft are isothermal. Exclusion of those situations 

 where the transmission anomaly curve is not a 

 straight line may be expected to give results system- 

 atically different from those obtained when runs are 

 classified only by the temperature distribution. 

 Therefore the average anomaly curves given below 

 are preferable as a tool for estimating the sound in- 

 tensities to be expected in any situation. 



Average Transmission Anomalies 



The average transmission anomalies obtained with 

 MIKE and CHARLIE patterns have been combined 



